基于随机共振的语音增强研究

Research on Speech Enhancement Based on Stochastic Resonance Mechanism

传统的语音增强方法是通过线性过滤除噪声达到语音增强的目的,而在强噪声环境中,语音信号表现为弱信号,去噪变得困难,而随机共振能利用噪声增强语音信号。基于Hodgkin-Huxley神经元阈上非周期随机共振原理,提出一种自适应调节,添加最佳噪声来实现语音增强。实验结果表明,在强噪声能够实现对语音信号的增强,具有一定的鲁棒性,提供了在强噪声环境中增强语音信号的新思路。

Speech enhancement is accomplished by linear filtering noise in the typical speech enhancement. In strong noises environment, signals perform as weak signals, so it is hard to eliminate noises. Based on theory of suprathreshold aperiodic stochastic resonance in Hodgkin-Huxley model, a new self-adaptive adjusting and adding optimum noise speech enhancement method is given. The results indicate that, in strong noises environment speech signals can be enhanced, and new method has characteristics of robustness. A new idea is given to enhance the speech signals in strong noises environment.